Reconstruction of Pt13 Clusters into Pt2(CO)m on CO Addition in NaY

Jan 21, 2009 - Indian Institute of Technology Bombay. , §. Current address: Max Planck Institute for Solid State Research, Heisenbergstrasse 1, D-705...
0 downloads 0 Views 1021KB Size
Download PDF
2352

J. Phys. Chem. C 2009, 113, 2352–2359

Reconstruction of Pt13 Clusters into Pt2(CO)m on CO Addition in NaY Zeolite Yasar Akdogan,† Sankaran Anantharaman,† Xiong Liu,†,§ Goutam Kumar Lahiri,‡ Helmut Bertagnolli,† and Emil Roduner*,† Institut fu¨r Physikalische Chemie, UniVersita¨t Stuttgart, Pfaffenwaldring 55 D-70569 Stuttgart, Germany, and Department of Chemistry, Indian Institute of Technology Bombay, 400076 Mumbai, India ReceiVed: August 25, 2008; ReVised Manuscript ReceiVed: NoVember 24, 2008

Pt13 clusters with extraordinarily high magnetization are interesting samples in the area of CO chemistry. For the first time paramagnetic platinum carbonyl samples were prepared and characterized by electron paramagnetic resonance (EPR) spectroscopy to improve the understanding of the CO adsorption effect on the structure of small Pt particles with X-ray absorption spectroscopy and Fourier transform infrared (FTIR) methods. In contrast to existing studies on the reduced Pt cluster-CO interaction in the literature, the structure of the paramagnetic Pt sample before CO adsorption is entirely known. The well-defined Pt13 cluster in NaY zeolite reconstructs into small Pt2(CO)m aggregates under CO atmosphere at room temperature. X-Band EPR spectroscopy shows that the multiplet spectrum of the Pt13 cluster at giso ) 2.35 disappears, and a new spectrum belonging to the Pt2(CO)m sample appears at giso ) 1.98. Extended X-ray absorption fine structure spectroscopy supports the Pt13 cluster destruction upon admission of CO and the formation of a dimeric platinum carbonyl complex with a nearest-neighbor coordination number that decreases from ca. 5.8 to 0.9. FTIR spectroscopy is a very sensitive technique for the C-O vibrational frequency. It reveals that Pt2(CO)m species have one type of linear and two types of bridge bonded CO groups. Initially, a weak band at 2108 cm-1 shows that Pt has a small positive charge, which is converting into zero charge upon CO desorption-adsorption cycles. The surprisingly sharp bands in the CO bridge bond region facilitate assignment, in contrast to the bands of CO bonds of other neutral platinum carbonyl samples. 1. Introduction Supported Pt catalysts are used in a large number of commercially important applications, including oxidation of CO and residual hydrocarbons in automotive exhausts, NO conversion, and hydrogenation in petrochemistry.1-5 Catalytic activity is strongly dependent on the catalyst particles and increases with decreasing cluster size because the reaction takes place on the surface of the catalyst. The dispersion, which designates the fraction of atoms at the surface, is higher for small particles compared to bigger ones. The distribution of cluster size is often not uniform, but some magic numbers are highly preferred and lead to the formation of closed-shell polyhedral structures with a particular stability.6 The highly symmetrical icosahedral structure with one of the smallest clusters matching a magic number of 13 can be synthesized in zeolite supercages by means of the so-called “ship-in-a-bottle” method. Recently, the Pt13 cluster formation in faujasite NaY zeolite following ion exchange, calcination, and reduction and its characterization in terms of electronic configuration, structure, and magnetization have largely been elucidated by X-band electron paramagnetic resonance (EPR), extended X-ray absorption fine structure (EXAFS), and superconducting quantum interference device (SQUID) measurements and supported by density functional theory (DFT) calculations.7,8 Carbon monoxide is one of the most frequently used probe molecules in catalysis to characterize supported metal particles. It can provide information on the size and composition of * Corresponding author. E-mail: [email protected]. † Universita¨t Stuttgart. ‡ Indian Institute of Technology Bombay. § Current address: Max Planck Institute for Solid State Research, Heisenbergstrasse 1, D-70569 Stuttgart, Germany.

clusters, and on the oxidation and coordination states of cations by using mainly Fourier transform infrared (FTIR), temperature programmed desorption (TPD), and EXAFS techniques.9-12 Platinum carbonyl samples can be considered as anionic or neutral clusters. It is known that anionic platinum carbonyl clusters called Chini complexes ([Pt3(CO)6]n2-, n ) 3-6) can be prepared from precursor complexes such as Pt(NH3)42+ or from calcined Pt2+ and CO,13,14 but neutral Pt carbonyl structures were obtained from reduced Pt particles and CO.15,16 The formation of the neutral platinum carbonyl cluster depends on the size of the starting reduced Pt cluster. For example, Pt particles with coordination number (CN) > 5 supported on SiO2 were stable under CO atmosphere, but the smaller Pt particles (CN < 5) in LTL zeolite reconstructed into small Pt carbonyl aggregates upon CO adsorption, as shown by EXAFS spectroscopy.16 Shifts in the frequency of the C-O stretching mode are often used to investigate the electronic properties and geometric structures of the Pt particles by using FTIR spectroscopy.16-18 The chemical bond between CO and Pt includes the 5σ Pt orbital, donating electron density from CO to Pt, and Pt 2π*, which transfers electron density back from Pt to an antibonding CO orbital. In general, two υ(CO) absorption regions around 2000 and 1850 cm-1 are assigned to linear and bridge bonding CO on Pt, respectively. Depending on the values and intensities of the absorption frequencies, bond formation is characterized by FTIR. Despite the general application of these methods to investigate metal clusters with adsorbed CO, the structures of the catalyst before and after CO exposure are still a matter of debate. The effect of CO adsorption on the structure of supported Pt particles is understood well when the structure of the Pt particles is known

10.1021/jp807566a CCC: $40.75  2009 American Chemical Society Published on Web 01/21/2009

Reconstruction of Pt13 Clusters into Pt2(CO)m

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2353

before and after CO adsorption. Therefore, in this study, we performed EPR spectroscopy, in situ X-ray absorption spectroscopy (XAS) combining X-ray absorption near edge structure (XANES) and EXAFS, and in situ FTIR techniques collectively. Taking advantage of the complementary information provided by these techniques to compare and characterize the Pt13/NaY cluster before and after CO exposure, the formation of smaller Pt2(CO)m aggregates from Pt13/NaY clusters was revealed. 2. Experimental Section Preparation of Pt13Hm Cluster. Pt/NaY samples were prepared by ion exchange with 3 mM [Pt(NH3)4]Cl2 solution to 1 g of NaY zeolite in 200 mL of water for 92 h at 343 K. The loading of Pt in NaY is 6 wt %, as verified by elemental analysis. Exchanged Pt(NH3)42+/NaY was calcined using a low ramping rate (0.5 K min-1) in flowing O2 (20 mL min-1 g-1) from room temperature to 573 K, and then holding at that temperature for 5 h. After cooling to room temperature in an oxygen flow, N2 was applied to purge the residual O2 in the reactor. The reduction was carried out in flowing hydrogen (25 mL min-1 g-1) with a heating rate of 6 K min-1 to 473 K for 1 h. The sample was cooled to room temperature under hydrogen flow, was transferred into EPR tubes in a glovebox under N2, and then evacuated. Deuterium exchanged samples were obtained after filling the EPR tube containing the Pt13Hm/NaY sample with 500 mbar of D2 for 2 days at room temperature. Evacuation of Pt13Hm samples at 408 K for 5 h using a turbomolecular pump yielded the hydrogen desorbed Pt13 cluster. XAS Spectroscopy. The XAS measurements were carried out at beamline X1 of Hasylab, Hamburg, Germany. A Si(111) double-crystal monochromator detuned to 65% of the intensity to remove higher harmonics was used for the energy scans. All spectra were recorded in transmission mode using ionization chambers filled with argon to achieve optimal absorption and signal-to-noise ratios. Energy calibration was performed using a platinum metal foil, which was measured simultaneously with every sample. The oxygen calcined Pt/NaY samples were pressed into pellets (13 mm diameter) and placed in a stainlesssteel in situ cell that was connected to a gas flow system. The heating and the temperature of the cell were regulated using a Eurotherm controller. XAS scans at the Pt LIII edge (11 564 eV) were performed at room temperature following the heating of the sample in a 25 mL min-1 flow of H2 at 473 K for 1 h, after argon gas flow at 50 mL min-1 for 10 min, and after CO (1 bar) gas flow at 25 mL min-1 for 20 min. The XAFS spectra measured on the energy scale are subjected to data evaluation, which involves background absorption correction, normalization, and background subtraction. In this work, the program AUTOBK implemented in the IFEFFIT19,20 program package was used for background correction, normalization, and background subtraction. The extracted EXAFS function was transformed to k-space by choosing a suitable threshold energy from half the absorption edge step. EXAFS data analysis was performed in k-space. The fit of the experimental data to the theory was obtained by adjustment of the common theoretical EXAFS expression according to the curved wave formalism of EXCURV98.21 The potential and phase shifts in this program were calculated employing Hedin-Lundqvist for the exchange potential and von Barth for the ground-state potential. The amplitude reduction factor was determined to be 0.8 in previous work7 and was fixed during the iteration procedure. When fitting experimental data with theoretical models, an inner potential correction Ef was introduced that accounts for an overall phase shift between the experimental

Figure 1. XANES spectra of Pt/NaY samples measured at the Pt LIII edge (11 564 eV) at different stages of in situ treatment.

and the calculated spectra. The quality of fit is given in terms of the R-factor, in accord with the literature.22 EPR Spectroscopy. Continuous wave (cw) X-band EPR spectra were recorded on a Bruker EMX spectrometer with a microwave frequency of about 9.466 GHz at 20 K. EPR spectral simulations were performed using WINEPR SimFonia, version 1.25. The spin concentration of a sample was calibrated against a standard sample (Ultramarine blue diluted by KCl) via double integration of the derivative signal. Pt13Dm/NaY (ca. 4 × 10-5 mol Pt) in an EPR tube was exposed to an excess amount of CO (natural isotopic abundance) or enriched CO (99 atom % 13C) gases at an initial 50 mbar pressure (ca. 10-4 mol) for 10 min at room temperature. The residual CO gas was evacuated over a few minutes from the EPR tube before the measurements. The next excess CO doses of higher pressures (100, 125, 500 mbar) were added by the same procedure. FTIR Spectroscopy. The FTIR spectra were obtained on a Magna-IR 560 spectrometer at a spectral resolution of 2 cm-1, accumulating 256 scans. Self-supporting pellets (10-20 mg cm-2) were prepared from the reduced Pt sample powder in air atmosphere. Additional reduction was applied after evacuation at 573 K for 2 h with 1.5 K min-1 heating rate to avoid oxidation and water vapor in the IR cell at 473 K for 1 h under a 100 mbar hydrogen atmosphere. The residual hydrogen was evacuated at room temperature for 1 h. Subsequently, the sample was exposed to 500 mbar of CO for 10 min and then evacuated for 3 min. 3. Results XANES. XANES spectra of the Pt metal foil, calcined Pt/ NaY, reduced Pt/NaY, reduced Pt/NaY after subsequent Ar purging, and CO adsorption on reduced Pt/NaY are given in Figure 1. Both the shape and the intensity of the white line23 of the XANES are found to be completely different, although the absorption edge positions of the different Pt samples are similar to those of the reference metal foil. Compared to the metal foil, the calcined sample exhibits a relatively intense white line which is found to decrease in intensity upon reduction and further upon Ar purging, but it is still larger than for the metal foil. Upon CO adsorption, there is again a strong increase in the white line intensity. EXAFS. The experimental and fitted EXAFS functions in k-space and the corresponding Fourier transforms in real space of the platinum clusters after in situ reduction are presented in Figure 2. The local structure around Pt in Pt13Hm/NaY obtained from the fit, the parameters used in the fitting procedure, the resulting fit parameters, and the estimated errors in their

2354 J. Phys. Chem. C, Vol. 113, No. 6, 2009

Figure 2. Experimental EXAFS function (top), its Fourier transform (bottom), and fit (red lines) to the experimental data of Pt13Hm/NaY.

determination are given in Table 1. In the experimental phase corrected Fourier transformed EXAFS function, one intense peak around 2.7 Å is observed along with a smaller side lobe at roughly 1.8 Å. It was verified by fitting that this is not a structure feature from lighter backscattering elements which might be present at this distance but is due to the nonlinear nature of the backscattering amplitude relevant in heavy elements. This feature is also reproduced when the theoretical EXAFS function was calculated using only Pt and oxygen shells at higher distances. In the EXAFS spectrum, only a single intense peak at roughly 2.7 Å is present and further peaks with comparable intensities, which are normally present in larger platinum clusters or bulk platinum metal, are absent. Due to this fact, the addition of higher shells of Pt in the fitting did not give a significant improvement of the fit. The peak near 2.7 Å could be fitted with Pt at 2.77 Å as well as O at 2.72 Å with average coordination numbers of 5.8 and 3.8, respectively. Considerable improvement in the quality of the fit justified the inclusion of this oxygen shell in the model along with the platinum shell. The first shell Pt-Pt average coordination number of 5.8 provides evidence that very small, probably 13 atom, clusters are formed in NaY. However, despite the small cluster size of ca. 0.8 nm, the observed Pt-Pt distance is comparable with its bulk value, ∼2.77 Å, due to a relaxation effect by the adsorbed hydrogen atoms. The presence of the longer Pt-O distance compared to that observed in PtO2, ∼2.0 Å, and the absence of an oxygen shell at this distance exclude the possible oxidation of the platinum clusters and reiterate the necessity of using in situ reduction to characterize such samples. The k2-weighted experimental EXAFS function k2χ(k), its Fourier transform for Pt2(CO)m/NaY (Figure 3), and their comparison with Pt13Hm/NaY are shown in Figure 4. Significant changes in the spectrum upon CO exposure at room temperature can be observed. Change in the backscattering amplitude, which is the envelope in the EXAFS function, is observed at both low and high k values. After CO adsorption, oscillations in the range 7-14 Å-1, relevant for heavy elements, are subdued, whereas strong oscillations appear in the range 3-7 Å-1. A considerable decrease in the intensity of the Fourier transform peak at roughly 2.7 Å and the appearance of peaks near 2.0 Å are noticed after CO adsorption. In order to quantify the observed changes,

Akdogan et al. experimental spectra were fitted with relevant structure models consisting of Pt-C, Pt-O, and Pt-Pt contributions. Based upon the goodness of fit, the most probable structure and eventually the local structure parameters such as average coordination number, average interatomic distances, and Debye-Waller-like factors for the different shells were obtained. For the CO adsorbed platinum cluster, the model consists of one Pt-Pt shell, two Pt-C shells, and one Pt-O shell (zeolite oxygen) contributions. The use of further oxygen shells from CO at more than 3 Å was statistically not justified due to the increase in the number of iteration parameters. The Pt-Pt distance in Pt2(CO)m/ NaY has contracted to 2.69 Å from 2.77 Å in Pt13Hm/NaY, together with a decrease in the Pt-Pt average coordination number from 5.8 to 0.9. Two Pt-C contributions, one at 1.95 Å and the next at 2.14 Å, are obtained with average coordination numbers of 0.8 and 1.7, respectively. X-Band EPR. Figure 5a shows an X-band EPR spectrum of reduced Pt in NaY that was measured at 20 K. The deuterium exchanged sample was used for EPR characterization because it gives a stronger and more symmetrical spectrum while the main feature of the spectra remains the same. The observed highly symmetrical multiplet has a giso value of 2.35 with a splitting which derives from 12 equivalent Pt nuclei. Adsorption of CO at room temperature on the reduced Pt cluster leads to a complete change of the EPR spectrum. The spectrum of the Pt cluster at giso ) 2.35 disappears, and a new spectrum appears at giso ) 1.98. The signal intensity of the newly formed carbonyl sample increases without any structural variations in the spectrum until 125 mbar of CO, but on further CO addition it has an almost constant intensity (Figure 5b-e). The double integral of the EPR spectra is proportional to the number of unpaired electrons, and after CO adsorption the resultant species has about 15 times more signal intensity than that of the reduced Pt cluster. Exposure of CO on the pure NaY (treated under the same condition of calcination and reduction) and on the Pt2+/NaY did not give any EPR signal. The EPR spectrum of the isotopically enriched sample of the 194Pt cluster/NaY facilitates the assignment of the platinum carbonyl structure. Figure 6 compares the spectra of the Pt cluster upon CO exposure composed of natural abundance Pt (33.3% 195Pt, 66.6% 194 Pt) and enriched 194Pt (97.5%). In contrast to the enriched 194 Pt sample (I ) 0) (Figure 6a), two small satellites on both sides of the main peak originate from hyperfine splitting of 195Pt nuclei (I ) 1/2) which contains Pt in natural abundance (Figure 6b). The enrichment is incomplete, so this small amount of Pt with I ) 1/2 causes the remaining weak hyperfine features seen in Figure 6a. FTIR. The FTIR spectra obtained after CO adsorption at room temperature under saturation conditions (PCO ) 500 mbar) on the reduced sample and then upon subsequent evacuation at the same temperature for different time durations are displayed in Figure 7. In order to avoid an exposure to air and, more importantly, to obtain measurement conditions similar to those used in the EPR measurements, the reduced Pt/NaY sample was subjected to additional treatments in the FTIR sample cell. The air effect can be thought of as a combination of oxygen and water vapor, which were studied previously by EPR.24 The EPR signal of the Pt13 cluster disappears upon interaction with oxygen or water vapor after long-time air exposure. On the other hand, after evacuation of the sample tube at 573 K for 2 h and upon reduction with static hydrogen pressure at 473 K, the Pt multiplet signal reappears. In the FTIR spectra, it is clear that the absorption around 2000 cm-1 is due to linearly coordinated CO and bands around

Reconstruction of Pt13 Clusters into Pt2(CO)m

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2355

TABLE 1: Fit Results of All EXAFS Data sample Pt13Hm Pt2(CO)m

Abs-Bsa

CN(Bs)b

R(Bs)c [Å]

σd [Å]

EFe [eV]

k-range [Å-1]

Pt-Pt Pt-O Pt-C Pt-C Pt-Pt Pt-O

5.8 ( 0.6 3.8 ( 0.4 0.8 ( 0.1 1.7 ( 0.2 0.9 ( 0.1 4.6 ( 0.7

2.77 ( 0.03 2.72 ( 0.03 1.95 ( 0.02 2.14 ( 0.02 2.69 ( 0.03 2.72 ( 0.03

0.084 ( 0.01 0.050 ( 0.01 0.055 ( 0.01 0.092 ( 0.01 0.059 ( 0.01 0.081 ( 0.01

-10.02

2.92-17.03

24.48

-13.44

2.90-17.01

34.73

a Absorber (Abs)-backscatterers (Bs). deviation. e Fermi energy EF.

b

Coordination number (CN).

c

Interatomic distance.

d

R-factor

Debye-Waller factor (σ) with its calculated

Figure 5. X-band EPR spectra of Pt13Dm recorded at 20 K before (a) and after adsorption of different doses of CO and evacuation at room temperature: 50 (b), 100 (c), 125 (d), and 500 mbar (e).

Figure 3. Experimental EXAFS function (top), its Fourier transform (bottom), and fit (red) to the experimental data of Pt2(CO)m/NaY.

Figure 6. EPR spectra of 194Pt2(CO)m (a) and natural abundance Pt2(CO)m (b) recorded at 20 K after saturation adsorption of CO at room temperature.

Figure 4. Experimental EXAFS function (top) and its Fourier transform (bottom) of Pt13Hm/NaY and Pt2(CO)m/NaY.

1850 cm-1 are caused by bridged CO coordinated to neighboring Pt atoms. The FTIR spectrum displays a peak at 2080 cm-1 with a shoulder at 2108 cm-1 which are assigned to bands of

linear Pt-CO (Figure 7). In the absorption region around the band of bridged Pt-CO, two clearly distinguishable peaks at 1893 and 1840 cm-1 with a shoulder at 1813 cm-1 are observed. Figure 7 shows the behavior of the absorption bands with different evacuation times at room temperature. CO evacuation leads to a decrease in the intensity of the peak at 2108, 1893, and 1840 cm-1, without shifting them to lower wavenumbers, and after 90 min these bands vanish. However, the strong band at 2080 cm-1 and the shoulder at 1812 cm-1 are not affected significantly. Two bridge bands at 1893 and 1840 cm-1 exhibit similar behavior upon evacuation. Figure 8 depicts variations of the FTIR spectra during the temperature programmed desorption. The bands at 2108, 1893, and 1840 cm-1 disappear at 373 K, but the main signal at 2080 cm-1 and the shoulder at 1812 cm-1 remain almost unchanged. At 473 K, the main signal at 2080 cm-1 shifts to 2070 cm-1 under formation of a new shoulder at 2000 cm-1, and the shoulder at 1812 cm-1 shifts to 1795 cm-1. At an even higher

2356 J. Phys. Chem. C, Vol. 113, No. 6, 2009

Akdogan et al. bands, the shoulder at 1813 cm-1 behaves like a new band and shows an increase in intensity after each cycle. As a result, the initial CO adsorbed-desorbed sample and the sample after three adsorption-desorption cycles exhibit spectra that are different from each other. 4. Discussion

Figure 7. FTIR spectra recorded after CO adsorption (500 mbar) at 298 K on Pt13Hm/NaY followed by evacuation at the same temperature for 3 (black), 25 (red), 45 (blue), and 90 min (green).

Figure 8. FTIR spectra recorded after CO adsorption (500 mbar) at 298 K on Pt13Hm/NaY followed by evacuation at 298 (a), 348 (b), 373 (c), 473 (d), and 523 K (e).

Figure 9. FTIR spectra of CO adsorbed on Pt13Hm/NaY at 500 mbar: (a) initial sample and after (b) 1-, (c) 2-, and (d) 3-fold CO adsorption-desorption cycles.

temperature of 523 K, the main band shifts to 2061 cm-1 with a noticeable lower intensity and an obvious shoulder at 2001 cm-1, along with the complete disappearance of the bridge bands. Figure 9 shows the FTIR spectra obtained after repeated room temperature 500 mbar CO adsorption-desorption cycles with the emphasis on the bands that occur at 2108, 2000, and 1813 cm-1. Following three adsorption-desorption cycles the shoulder at 2108 cm-1 nearly disappears, but the one at 2000 cm-1 becomes more pronounced. In contrast to the behavior of other

XANES. The XANES shows the difference in the shape and height of the white line in the case of the sample prepared by oxygen calcination prior to in situ reduction, after hydrogen reduction, and after subsequent argon purging and CO adsorption. The LIII edge in platinum corresponds to the electronic transition from the 2p3/2 to the 5d level.25,26 Platinum has the electronic configuration 5d96s1 in its ground state, and the unoccupied d level in the Pt atom is the d5/2 state. Depending upon the d-band vacancies in the final state, the intensity of the white line varies. After calcination, the Pt has a higher oxidation state with more vacancies, which causes an intense white line. Increase in the white line intensity for hydrogen-reduced small Pt clusters compared to the platinum metal foil has previously been attributed to changes in the d-band vacancies.27 However, in this case, both the effect of the cluster size and the effect of hydrogen adsorption play a role. In order to decouple these two effects, one has to measure the XANES for bare clusters and hydrogen adsorbed clusters. In this respect, upon Ar purging (Figure 1), reduction in the intensity and width of the white line is observed. This tendency can be attributed to the partial desorption of adsorbed hydrogen, though complete desorption of hydrogen could not be verified. It should be noted that the Pt-H bonds are strongly polar, with the negative end on the hydridic hydrogen, which partly depletes the d state. Upon CO adsorption, the increase in the white line could be attributed to the increase in ionicity of the platinum center due to bond formation with CO. Thus, the platinum atoms have a lower electron density by π-back-donation from Pt to CO or a positive partial charge on Pt, or a combination of both.15 Qualitatively, on moving from the metal foil to small clusters in the zeolite with adsorbed hydrogen or CO, the changes in the white line are attributed to an increase in d-band vacancies. EXAFS. EXAFS measurement and analysis of reduced Pt samples in NaY zeolite prepared ex situ using similar conditions has already been reported in an earlier work.7 In this work, the Pt cluster preparation has been carried out under in situ conditions under a high flux of the reactant gases so that air oxidation of the clusters can be safely ruled out. Additionally, the effect of subsequent CO adsorption on the structure of platinum clusters has been investigated. Insights into the size of the clusters, their location in the zeolite matrix, and the effect of CO adsorption were obtained from the analysis of the local structure around Pt. The average Pt-Pt shell coordination number of 5.8 obtained from the analysis is compatible with the presence of 13 atom clusters, also evidenced by EPR, which is discussed in the following section. The clusters of this size (ca. 0.8 nm) could very well fit into the supercages of the NaY zeolite with a free diameter of 1.3 nm. In addition, a second near neighbor shell contribution from the zeolite oxygen atoms is present at 2.72 Å. Similar Pt-O distances have already been reported earlier for Pt in Y zeolites.28 The relatively long Pt-O distance is attributed to the presence of interfacial hydrogen after reduction.16 Analysis of Pt13Hm after CO adsorption revealed a contraction of the Pt-Pt distance from 2.77 to 2.69 Å, accompanied by a decrease in the Pt-Pt coordination number from 5.8 to 0.9. This average Pt-Pt distance is very similar to that measured for

Reconstruction of Pt13 Clusters into Pt2(CO)m [Pt2Br2(µ-CO)(PPh3)] compounds with 2.65 Å.29 The changes in the structure parameters indicate a reconstruction of the Pt13 cluster into smaller aggregates, most likely Pt2. Furthermore, the two different Pt-C contributions, one at 1.95 Å and the other at 2.14 Å with coordination numbers of 0.8 and 1.7, respectively, are obtained. The short Pt-C distances are similar to Pt-C distances observed in compounds like Pt3(CO)6 where CO molecules are linearly and bridge coordinated to Pt.30 Therefore, it can be said with some degree of certainty that, in the present case, one linear and two bridged CO molecules coordinate the Pt atoms. The presence of two different Pt-C contributions is attributed to the Pt-CO linear and bridge bonded species as confirmed by the FTIR results. A further EXAFS contribution is present at 2.72 Å due to zeolite oxygen with an increase in coordination number of Pt-O from 3.8 to 4.6, indicating that the Pt carbonyl clusters are in proximity to the zeolite framework. This observation is attributed to the stabilizing effect of the zeolite oxygen atoms on the resulting structure. In this case, neutral platinum carbonyl samples having only CO ligands are very unstable due to weak Pt-CO bonds, and so far only Pt(CO)4 has been prepared in argon matrixes below 10 K.31,32 However, oxygen atoms in KL zeolite acted as σ-donor ligands to stabilize the small platinum carbonyl clusters.18 The basicity strength of the oxygen atom in NaY is similar to that in KL zeolite,33 so it is reasonable to obtain small stable platinum carbonyl species closer to the framework oxygen. Furthermore, we have limited information about the residual charge on the carbonyl cluster (see the FTIR section below). EPR. Pt clusters were studied extensively as hydrogen adsorbed Pt13Hm, deuterium exchanged Pt13Dm, and as bare Pt13 clusters by X-band EPR in a previous study.8 The well-defined paramagnetic cluster containing 12 equivalent surface atoms and a single core atom has a multiplet spectrum at giso ) 2.35 with a splitting due to 12 equivalent Pt nuclei. Pt at the core of the cluster does not carry significant unpaired electron density since for symmetry reasons the relevant cluster molecular orbitals have a nodal plane through the center. The highly symmetrical EPR spectra suggest an icosahedral structure in the supercages of the NaY zeolite, although open shell structures are expected to undergo a Jahn-Teller distortion. CO adsorption dissipates the signal of the Pt13 cluster, and this was not reversible even on desorption of CO and subsequent hydrogen reduction. It indicates that the Pt13 clusters are completely destroyed after CO reaction (Figure 5). The shape of the newly formed Pt-CO spectrum is retained on increasing intensity upon additional CO exposure, which means that only one type of paramagnetic Pt carbonyl species is formed. Moreover, only one pattern of g parameters reveals that the location of this complex within the zeolite seems to be the same. The 15-fold higher yield calculated from the double integral of the platinum carbonyl spectrum compared with that of reduced Pt13 clusters shows that the number of unpaired electrons is increasing 15 times in the Pt2(CO)m/NaY sample. However, the amount of EPR active Pt is only 0.5%, which means that most of the Pt is still EPR silent. The rest can be diamagnetic platinum carbonyl or found in high spin states, which are not accessible to X-band EPR but can be studied by SQUID measurements. The new spectrum at giso ) 1.982 is composed of two satellites on both sides of the main signal, but the enriched 194 Pt-CO sample shows only the main signal. The predicted hyperfine multiplet arising from two Pt nuclei with spin 1/2 with 1:8:18:8:1 relative intensities is very compatible with the

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2357

Figure 10. EPR spectra of Pt2(CO)m/NaY sample recorded at 20 K and corresponding simulation.

TABLE 2: EPR Parameters from Spectral Simulation Obtained with Pt Samples A(195Pt) [MHz]

g samples

g1

g2

g3

g⊥

g|

A1

A2

A3

A⊥

A|

Pt13Dm 2.356 2.359 234 211 Pt2(CO)m 1.957 1.985 2.005 290 222 235 194 Pt2(CO)m 1.954 1.987 1.989

experimental Pt-CO EPR spectrum. The EPR spectrum of the saturated carbonyl species (Figure 5e) and the corresponding simulation are displayed in Figure 10. The assigned structure is a Pt dimer which is composed either of two, one, or zero I ) 1/2 nuclei (195Pt), the remaining nuclei being of spin zero, with natural abundance of isotopes. The g and A tensors are listed in Table 2. The g tensor of the Pt dimer is rhombic with a small anisotropy, g1 ) 1.957, g2 ) 1.985, and g3 ) 2.005, indicating that the axial symmetry is broken by side-on coordination at a cage wall or by the coordinated CO molecules. Pt dimers have been observed elsewhere in NaA zeolite, and others in single crystal silicon with an average g value of 2.30 and components ranging between 1.78 and 2.17, respectively.9,34-36 Furthermore, an identical EPR spectrum is obtained when 13CO (carbon: I ) 1/2, 99%) is used instead of natural abundance carbon (I ) 1/2, 1.1%; I ) 0, 98.9%) without additional hyperfine lines or line broadening due to 13C interaction. This indicates that the unpaired spin density is mainly located on Pt atoms and it is almost zero at the 13C nucleus. The isotropic hyperfine interaction (A1 + A2 + A3)/3 is about 250 MHz, and for 195Pt atoms the tabulated value for Aiso is +34 410 MHz, so the fractional 6s contribution is 0.7% per Pt atom. The experimental anisotropic hyperfine interaction (A1 - (A2 + A3)/2)/3 amounts to 21 MHz, and the corrected tabulated Aaniso value is 2/7 × 1474 ) 421 MHz for the 5d orbital.37 If all hyperfine components are positive, the estimated 5d contribution is about 5% per Pt atom. Such low 5d contributions can be explained if one or two of the hyperfine components are of opposite signs, for example, this yields 23% or 41% 5d character per Pt atom by assuming that A3 is negative, or A2 and A3 are both negative, respectively. FTIR. FTIR spectroscopy has been used to characterize the reactivity of small supported Pt clusters with CO by observing the C-O stretching frequency, which depends on the Pt atom to which CO is bound. The C-O vibration is very sensitive to the number of Pt atoms linked to CO ligands,38 the coordination and oxidation state of the Pt atom to which CO is adsorbed,9,39 the CO coverage,40 and the acid-base properties of the support.41 Moreover, the intensity ratio of the bands of linear

2358 J. Phys. Chem. C, Vol. 113, No. 6, 2009 and bridge bonded CO is influenced by the electron charge on the supported Pt cluster.41 The FTIR experiments that were carried out by considering these factors show that the dimer platinum carbonyl sample has one type of linear bonded and two different bridge bonded CO molecules. Since the structures of bare and hydrogen covered Pt13 clusters are the same and the adsorbed hydrogen atoms are easily exchanged with CO molecules, the IR spectra after CO adsorption are similar over both platinum samples. The absorption band at 2080 cm-1 with a small bandwidth (half-bandwidth of 44 cm-1) at room temperature can be assigned to linearly bound Pt0-CO species; however, the shoulder at 2108 cm-1 is different from the main band and behaves independently. Carbonyl bands at frequencies >2100 cm-1 have usually been assigned to Ptδ+-CO species.9,42 The EPR active fraction of the precursor Pt13Hm cluster probably has an odd positive charge (+1 or +3),8 and this may be retained upon admission of CO under fragmentation to Pt-CO aggregates to preserve the charge balance in zeolite. The major fraction is EPR inactive and probably a high spin state of unknown charge, but EXAFS shows that these clusters are of the same size. Since CO is a reducing agent, the charge of Pt in the newly formed platinum carbonyl species must be close to zero or adopt at best a small positive value. Therefore, the absorption at 2108 cm-1 can be assigned to Pt2δ+-CO, with slightly electron deficient Pt atoms. In concentrated sulfuric acid, the homoleptic, dinuclear, cationic platinum(I) carbonyl complex, [{Pt(CO)3}2]2+, was synthesized from PtO2 by exchange with CO.43 This shows that it is possible to stabilize also cationic platinum carbonyls in suitable media. By evacuation at room temperature or at high temperatures, the corresponding band easily disappears, which shows that the Ptδ+-CO bond is weak due to low Pt f π* back-donation. After two adsorption-desorption cycles, this band almost did not reappear after the addition of the next 500 mbar CO aliquot (Figure 9). This implies that either Ptδ+ reduces to Pt0 or that new Pt carbonyl species are present. The presence of both of these is likely because the Pt20-CO band intensity increases and new signals at 2001 and 1813 cm-1 become more prominent. According to EXAFS and in agreement with EPR, the new unit formed on fragmentation is binuclear and contains a Pt-Pt bond. The question is how many CO ligands it contains. The bands due to bridged CO at 1893 and 1840 cm-1 change in concert, which shows that they relate to the υs and υas modes of two coupled CO ligands of the same species. These bands are very similar to bridge bands belonging to subcarbonyl species of Chini complexes, e.g., [Pt3(CO)3(µ2-CO)3], or triangular cluster framework species at 1896 and 1841 cm-1 after reductive carbonylation of Pt2+/NaY upon exposure to CO at room temperature.13 This implies that a symmetric bridge bonded structure is formed. Additionally, one more type of bridged CO group is observed at 1813 cm-1, especially after adsorptiondesorption cycles. Partial desorption of CO leads to a change in the position of the platinum carbonyls and allows the cluster to be closer to the zeolite framework, which causes limitation of free space for CO bridge bonding. Figure 9 shows a huge increase in intensity of the band at 1813 cm-1, which is now more favorable, compared to that of other types of bridged CO bands. There are one linear and two bridge bonded CO ligands (sym, asym) per Pt atom at the initial coverage, in agreement with the EXAFS results which give one carbon at a distance of 1.95 Å and two at 2.14 Å. After the desorption-adsorption cycles the band of another bridged CO ligand (1813 cm-1) is

Akdogan et al.

Figure 11. Schematic structure of Pt2(CO)m along with possible types of CO bonding.

more pronounced. We therefore suggest that the dinuclear Pt carbonyl complex has the structure given in Figure 11. The cohesive energy per atom scales approximately with the average coordination number.44 With 347 kJ mol-1 as calculated for 13-atom metal clusters (CN ) 5.8 ( 0.6, see Table 1),8 it is considerably smaller than the 528 kJ mol-1 for the bulk metal (CN ) 12).45 For Pt10 clusters the calculated average cohesive energy per Pt atom is 255 kJ mol-1, and the heat of formation of neutral Pt-CO clusters by reaction of Pt10 with 10 CO molecules was calculated to be only -30 kJ/mol CO, which is nearly thermoneutral.45 Therefore, on CO addition small particles are highly reactive toward fragmentation, since the metalcarbonyl bonds compensate for the lost cohesive energy. For example, when a Rh/Al2O3 catalyst was studied before and after CO adsorption, it was found that CO adsorption disrupts the Rh-Rh bonds and decreases the Rh-Rh coordination number from 3.7 to 1.8.46 On the other hand, rhodium particles with an average coordination number CN ∼ 7 did not reconstruct under CO atmosphere.47 Here, EXAFS and FTIR studies have shown that the Pt metal particle size effect is similar to that of Rh with CO.15,16 However, the FTIR spectra of a CO exposed Pd cluster in NaY was assigned to Pd13(CO)x clusters. It was proposed that the Pd13 cluster is obtained in the supercages and that it is stable on CO adsorption, although the particle size is not very large.48 The interpretation of such experimental data is often ambiguous because the EXAFS technique gives average fit values, and a frequency shift of CO is usually the result of more than one effect. In order to clarify the size effect, the powerful technique of EPR spectroscopy should not be underestimated. 5. Conclusion Small monodisperse Pt13 clusters in the NaY zeolite are very reactive toward CO at room temperature. CO adsorption leads to a complete decomposition of the cluster and to the formation of new Pt2(CO)m (m ) 4-5) species, most probably stabilized by the zeolite walls. Paramagnetic properties of these Pt samples are utilized to investigate the structure of the small clusters by using EPR hyperfine structure. It yields results which are consistent with in situ EXAFS measurements. The number of bonded carbon monoxides on a per Pt atom basis is difficult to obtain, but the analysis of C-O stretching frequencies helped in determining one type of linearly adsorbed CO at 2080 cm-1, and two bridge bonded COs at υs ) 1893 cm-1 and υas ) 1840 cm-1, which is again compatible with EXAFS results. In addition, a further bridge bonded species at 1813 cm-1 is preferentially formed during desorption-adsorption cycles. A cationic platinum carbonyl species has also been seen with a band at 2108 cm-1, but the main complex has an absorption band at 2080 cm-1, belonging to the neutral Pt20-CO structure.

Reconstruction of Pt13 Clusters into Pt2(CO)m This study also demonstrates the use of the various complementary methods to understand the structure of small clusters. Although EPR spectroscopy is not yet used extensively in the field of catalysis, paramagnetism offers a good possibility to follow chemical reactions and to understand the geometry of the species. Acknowledgment. S.A. wishes to thank Eric Wendel and the mechanical workshop for the construction of the in situ XAS cell, and Dr. Adam Webb from X1 Beamline at HASYLAB for his support during the XAS measurements. Financial support by the Deutsche Forschungsgemeinschaft via the SFB 706 and the Research Training Groups “Advanced Magnetic Resonance Type Methods in Materials Science” (GK448) and “Chemistry in Interphases” (GK441) is gratefully acknowledged. G.K.L. thanks the Deutsche Forschungsgemeinschaft for a Mercator visiting professorship. References and Notes (1) Burch, R.; Breen, J. P.; Meunier, F. C. Appl. Catal., B: EnViron. 2002, 39, 283–303. (2) Ciccariello, S.; Benedetti, A.; Pinna, F.; Strukul, G.; Juszczyk, W.; Brumberger, H. Phys. Chem. Chem. Phys. 1999, 1, 367–372. (3) Sachtler, W. M. H. Catal. Today 1992, 15, 419–429. (4) Tzou, M. S.; Teo, B. K.; Sachtler, W. M. H. J. Catal. 1988, 113, 220–235. (5) Zheng, J.; Dong, J. L.; Xu, Q. H. Stud. Surf. Sci. Catal. 1994, 84, 1641–1647. (6) Roduner, E. Chem. Soc. ReV. 2006, 35, 583–592. (7) Liu, X.; Bauer, M.; Bertagnolli, H.; Roduner, E.; Van Slageren, J.; Phillipp, F. Phys. ReV. Lett. 2006, 97, 253401-06. (8) Liu, X.; Dilger, H.; Eichel, R. A.; Kunstmann, J.; Roduner, E. J. Phys. Chem. B 2006, 110, 2013–2023. (9) Akdogan, Y.; Vogt, C.; Bauer, M.; Bertagnolli, H.; Giurgiu, L.; Roduner, E. Phys. Chem. Chem. Phys. 2008, 10, 2952–2963. (10) Kappers, M. J.; Miller, J. T.; Koningsberger, D. C. J. Phys. Chem. 1996, 100, 3227–3236. (11) Mojet, B. L.; Kappers, M. J.; Miller, J. T.; Koningsberger, D. C. 11th International Congress on Catalysiss40th AnniVersary, Parts A and B; Studies in Surface Science and Catalysis 101; Elsevier: Amsterdam, 1996; pp 1165-1174. (12) Van Santen, R. A. J. J. Chem. Soc., Faraday Trans. 1987, 8, 1915– 1934. (13) Li, G. J.; Fujimoto, T.; Fukuoka, A.; Ichikawa, M. Catal. Lett. 1992, 12, 171–186. (14) Longoni, G.; Chini, P. J. Am. Chem. Soc. 1976, 98, 7225–7231. (15) Mojet, B. L.; Koningsberger, D. C. Catal. Lett. 1996, 39, 191– 196. (16) Mojet, B. L.; Miller, J. T.; Koningsberger, D. C. J. Phys. Chem. B 1999, 103, 2724–2734. (17) Kustov, L. M.; Ostgard, D.; Sachtler, W. M. H. Catal. Lett. 1991, 9, 121–126.

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2359 (18) Stakheev, A. Y.; Shpiro, E. S.; Jaeger, N. I.; Schulz-Ekloff, G. Catal. Lett. 1995, 34, 293–303. (19) Newville, M. J. Synchrotron Radiat. 2001, 8, 322–324. (20) Ravel, B.; Newville, M. J. Synchroton Radiat. 2005, 12, 537–541. (21) Gurman, S. J.; Binsted, N.; Ross, I. J. Phys. C: Solid State Phys. 1984, 17, 143–151. (22) Binsted, N. In EXCURV98: CCLRC Daresbury Laboratory computer program manual; CCLRC: Daresbury, 1998. (23) Brown, M.; Peierls, R. E.; Stern, E. A. Phys. ReV. B 1977, 15, 738–744. (24) Liu, X. Ph.D. Thesis, Stuttgart University, 2006. (25) Koningsberger, D.; Mojet, B.; Miller, J.; Ramaker, D. J. Synchrotron Radiat. 1999, 6, 135–141. (26) Samant, M. G.; Boudart, M. J. Phys. Chem. 1991, 95, 4070–4074. (27) Koningsberger, D. C.; Oudenhuijzen, M. K.; Bitter, J. H.; Ramaker, D. E. Top. Catal. 2000, 10, 167–177. (28) Ji, Y. Y.; Van der Eerden, A. M. J.; Koot, V.; Kooyman, P. J.; Meeldijk, J. D.; Weckhuysen, B. M.; Koningsberger, D. C. J. Catal. 2005, 234, 376–384. (29) Goodfellow, R. J.; Herbert, I. R.; Orpen, A. G. J. Chem. Soc., Chem. Commun. 1983, 23, 1386–1388. (30) Torigoe, K.; Remita, H.; Picq, G.; Belloni, J.; Bazin, D. J. Phys. Chem. B 2000, 104, 7050–7056. (31) Doerr, M.; Frenking, G. Z. Anorg. Allg. Chem. 2002, 628, 843– 850. (32) Ku¨ndig, E. P.; Mcintosh, D.; Moskovit, M.; Ozin, G. A. J. Am. Chem. Soc. 1973, 95, 7234–7241. (33) Weitkamp, J.; Puppe, L. Catalysis and Zeolites; Springer: Berlin, 1999. (34) Hohne, M. Phys. ReV. B 1992, 45, 5883–5886. (35) Vonbardeleben, H. J.; Stievenard, D.; Brousseau, M.; Barrau, J. Phys. ReV. B 1988, 38, 6308–6311. (36) Ishiyama, T.; Fukuda, T.; Yamashita, Y.; Kamiura, Y. Physica B 2006, 376, 89–92. (37) Morton, J. R.; Preston, K. F. J. Magn. Reson. 1978, 30, 577–582. (38) Sheppard, N.; Nguyen, T. T. AdV. Infrared Raman Spectrosc. 1978, 5, 67. (39) Kappers, M. J.; Vandermaas, J. H. Catal. Lett. 1991, 10, 365–373. (40) Blyholder, G. J. Phys. Chem. 1975, 79, 756–761. (41) Visser, T.; Nijhuis, T. A.; Van der Eerden, A. M. J.; Jenken, K.; Ji, Y. Y.; Bras, W.; Nikitenko, S.; Ikeda, Y.; Lepage, M.; Weckhuysen, B. M. J. Phys. Chem. B 2005, 109, 3822–3831. (42) Chakarova, K.; Hadjiivanov, K.; Atanasova, G.; Tenchev, K. J. Mol. Catal. A: Chem. 2007, 264, 270–279. (43) Xu, Q.; Souma, Y.; Heaton, B. T.; Jacob, C.; Kanamori, K. Angew. Chem., Int. Ed. 2000, 39, 208–209. (44) Roduner, E. Nanoscopic Materials Size-Dependent Phenomena; RSC: Cambridge, 2006. (45) Watwe, R. M.; Spiewak, B. E.; Cortright, R. D.; Dumesic, J. A. Catal. Lett. 1998, 51, 139–147. (46) Vantblik, H. F. J.; Vanzon, J. B. A. D.; Huizinga, T.; Vis, J. C.; Koningsberger, D. C.; Prins, R. J. Am. Chem. Soc. 1985, 107, 3139–3147. (47) Vantblik, H. F. J.; Vanzon, J. B. A. D.; Koningsberger, D. C.; Prins, R. J. Mol. Catal. 1984, 25, 379–396. (48) Sheu, L. L.; Kno¨zinger, H.; Sachtler, W. M. H. J. Am. Chem. Soc. 1989, 111, 8125–8131.

JP807566A